Recent advancements in deep learning have produced significant progress in markerless human pose estimation, making it possible to estimate human kinematics from single camera videos without the need for reflective markers and specialized labs equipped with motion capture systems. Such algorithms have the potential to enable the quantification of clinical metrics from videos recorded with a handheld camera. Here we used DeepLabCut, an open-source framework for markerless pose estimation, to fine-tune a deep network to track 5 body keypoints (hip, knee, ankle, heel, and toe) in 82 below-waist videos of 8 patients with stroke performing overground walking during clinical assessments. We trained the pose estimation model by labeling the keypoints in 2 frames per video and then trained a convolutional neural network to estimate 5 clinically relevant gait parameters (cadence, double support time, swing time, stance time, and walking speed) from the trajectory of these keypoints. These results were then compared to those obtained from a clinical system for gait analysis (GAITRite®, CIR Systems). Absolute accuracy (mean error) and precision (standard deviation of error) for swing, stance, and double support time were within 0.04 ± 0.11 s; Pearson’s correlation with the reference system was moderate for swing times (r = 0.4–0.66), but stronger for stance and double support time (r = 0.93–0.95). Cadence mean error was −0.25 steps/min ± 3.9 steps/min (r = 0.97), while walking speed mean error was −0.02 ± 0.11 m/s (r = 0.92). These preliminary results suggest that single camera videos and pose estimation models based on deep networks could be used to quantify clinically relevant gait metrics in individuals poststroke, even while using assistive devices in uncontrolled environments. Such development opens the door to applications for gait analysis both inside and outside of clinical settings, without the need of sophisticated equipment.
22 April 2025 in Clinical Neurophysiology Practice
With growing interest in translating transcutaneous spinal cord stimulation (tSCS) into rehabilitation for different neurologic injuries, understanding the effects of various combinations of stimulation parameters becomes essential. Twenty-one participants post-stroke completed an assessment to determine their resting motor threshold (RMT) (minimum current required to elicit a muscle response) and tolerance levels (uncomfortable current intensity) to 12 stimulation configurations: two square waveforms, biphasic and monophasic, paired with six carrier frequencies (unmodulated: 0, and modulated: 1, 3, 5, 7, and 10 kHz). The results demonstrated that increasing carrier frequency increased participants' tolerance level and RMTs. Carrier frequency nor waveform type significantly altered discomfort when tolerance was normalized to the motor threshold, with 57 ± 23 % tolerated across all configurations. However, higher carrier frequencies, particularly biphasic waveforms at frequencies > 5 kHz, required more charge to reach a muscle activation and activated fewer muscles compared to unmodulated waveforms. No significant differences in discomfort relative to RMT were found between monophasic and biphasic waveforms. Higher carrier frequency allows stimulation to be more comfortable at a given intensity, but it also requires more current to reach RMTs. This study provides an essential feasibility assessment of tSCS configurations in a neurological population.
An increasing number of studies suggests that a novel neuromodulation technique targeting the spinal circuitry enhances gait rehabilitation, but research on its application to stroke survivors is limited. Therefore, we investigated the characteristics of spinal motor-evoked responses (sMERs) from lower-limb muscles obtained by transcutaneous spinal cord stimulation (tSCS) after stroke compared to age-matched and younger controls without stroke. Thirty participants (ten stroke survivors, ten age-matched controls, and ten younger controls) completed the study. By using tSCS applied between the L1 and L2 vertebral levels, we compared sMER characteristics (resting motor threshold (RMT), slope of the recruitment curve, and latency) of the tibialis anterior (TA) and medial gastrocnemius (MG) muscles among groups. A single pulse of stimulation was delivered in 5 mA increments, increasing from 5 mA to 250 mA or until the subjects reached their maximum tolerance. The stroke group had an increased RMT (27–51%) compared to both age-matched (TA: p = 0.032; MG: p = 0.005) and younger controls (TA: p < 0.001; MG: p < 0.001). For the TA muscle, the paretic side demonstrated a 13% increased latency compared to the non-paretic side in the stroke group (p = 0.010). Age-matched controls also exhibited an increased RMT compared to younger controls (TA: p = 0.002; MG: p = 0.007), suggesting that altered sMER characteristics present in stroke survivors may result from both stroke and normal aging. This observation may provide implications for altered spinal motor output after stroke and demonstrates the feasibility of using sMER characteristics as an assessment after stroke.
This is the first study to examine how electrode configuration influences stimulation comfort in transcutaneous spinal cord stimulation (tSCS) relative to resting motor threshold (RMT) in individuals with stroke. Across five configurations, most were tolerated at ∼50% RMT, whereas a two-channel setup reduced tolerance (∼35%). Muscle selectivity did not differ significantly among configurations. These findings establish critical feasibility benchmarks for electrode placement in future tSCS protocols for stroke rehabilitation.
1 October 2019 in Archives of Physical Medicine and Rehabilitation